Retrospective Theses and Dissertations Iowa State University Capstones, Theses andDissertations

1964

E.M.F. study of lower oxidation states in moltensodium tetrachloroaluminateTheodore Charles Fegley MundayIowa State University

Follow this and additional works at: https://lib.dr.iastate.edu/rtd

Part of the Inorganic Chemistry Commons

This Dissertation is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State UniversityDigital Repository. It has been accepted for inclusion in Retrospective Theses and Dissertations by an authorized administrator of Iowa State UniversityDigital Repository. For more information, please contact digirep@iastate.edu.

Recommended CitationMunday, Theodore Charles Fegley, "E.M.F. study of lower oxidation states in molten sodium tetrachloroaluminate" (1964).Retrospective Theses and Dissertations. 3871.https://lib.dr.iastate.edu/rtd/3871

This dissertation has been 65—4627 microfilmed exactly as received

MUNDAY, Theodore Charles Fegley, 1937-E.M.F. STUDY OF IJOWER OXIDATION STATES IN MOLTEN SODIUM TETRA-CHIJOROALUMINATE.

Iowa State University of Science and Technology, Ph.D., 1964 Chemistry, inorganic

University Microfilms, Inc., Ann Arbor, Michigan

E.M.F. STUDY OF LOWER OXIDATION STATES IN MOLTEN

Theodore Charles Fegley Munday

A Dissertation Submitted to the

Graduate Faculty in Partial Fulfillment of

The Requirements for the Degree of

DOCTOR OF PHILOSOPHY

Major Subject: Inorganic Chemistry

SODIUM TETRACHLOROALUMINATE

by

Approved:

I ^ Work

Iowa Iowa State University Of Science and Technology

Ames, Iowa

1964

Signature was redacted for privacy.

Signature was redacted for privacy.

Signature was redacted for privacy.

11

TABLE OF CONTENTS

Page

DEDICATION ill

INTRODUCTION I

EXPERIMENTAL DETAILS 10

General Cell Design 10

Cell Components and Construction 12

Experimental Procedure 19

Materials 22

Other Equipment 27

RESULTS AND DISCUSSION 30

Data Treatment 30

Initial Impurities 34

Cadmium 37

Lead 44

Tin 49

Zinc 53

Other Experiments 54

Proposed Research 57

SUMMARY 63

BIBLIOGRAPHY 66

ACKNOWLEDGMENTS 70

APPENDIX 71

ill

DEDICATION

To my father and mother

1

INTRODUCTION

In recent years much has been learned about the solution

of metals in their molten salts. Many of the observed phe­

nomena have been successfully correlated and interpreted by

the theory which proposes lower oxidation states. This

explanation holds that solution takes place by reactions

between the metal and its ions in a normal oxidation state in

the melt to yield one or more lower oxidation states. Thus,

the nature of the lower oxidation state species is and has

been of interest.

The lower oxidation state explanation has served espe­

cially well for the solution of post-transition metals in

their molten halides. For many of the elements, the stabil­

ity of lower states is large enough under certain conditions

to permit isolation of discrete solids. Besides the common

mercury(l) and thallium(l) compounds, solid, lower state

phases of cadmium [Cd2(AlCl^^2 (1)3, gallium [Gal (2) and

GaAlCl^ (3)], indium [InAlCl^ (4)], and bismuth [BiCl^

(5, 6)] have been isolated. In addition, a substance of

composition Al]_ 22^ was prepared and interpreted to be a

mixture of aluminum and All (7).

For cases where such reduced solids have been found,

2

the existence of the same lower state as a melt species

would seem probable. For instance, a wide variety of solu-

2 + tion studies point to the existence of Cd^ (8) in the

Cd-CdCl2 system where no reduced solid exists. Definitive

evidence for melt species related to the stoichiometry of

isolated solids is limited to Raman spectra of melts which

first, show Cd^"*" in molten Cd2(AlCl^)2-Cd(AlCl^)2 (9) and

second, eliminate all but Ga"*" in molten "gallium dichloride"

(10) .

The isolated reduced solid Al^ 22^ (7), gas phase

spectra for the Al-AlCl^ system (11), and calculations on

All which agree with observed solubilities in the AI-AII3

system (12) all attest to the probable existence of Al"*" in

melts. Nevertheless, direct experiments on the melts them­

selves are still lacking. In"*" remains consistent with

studies of the In-InClg (13-15), InCl-InCl^ (4), and InCl-

AlClg (4) phase diagrams, but compounds such as In^Cl^,

In^Clg, In^Cly, In^Clg are reported as well as InCl, so

complex species may also be present in addition to the simple

ion, ln+. Evidence for two reduced states of bismuth in

Bi-BiX^ melts does not include the Big"*" unit present in the

solid BiCl^ (8).

3

Separation of solid subsalts of the remaining post-

transition metals has not been achieved. Studies of the

Zn-ZnX2 systems are scarce and do not give any clear indica­

tion of what reduced species is in the melt. Topol (16)

attempted e.m.f. measurements on the Zn-ZnCl2 system, but was

not successful because potentials were unsteady. The exist­

ence of a lower oxidation state of tin is suggested by its

small solubility in molten SnCl2 and is supported by Russian

polarographic (17), chronopotentiometric (18), and electro-

deposition (19, 20) observations, but a specific species has

not been claimed.

Early e.m.f. work by Karpachev, et al. (21) suggested

a Pb+ solute in Pb-PbCl2 at 700°, but more recent determina­

tions by the same method conclude that Pb^^ is present in the

Pb-Pbl2 system at 585 and 693°, and is probably present in

Pb-PbCl2 melts (16). Egan (22) also arrived at Pb^^ by

polarographic determination of the concentration of the

reduced species in PbCl2 equilibrated with Fb-Au alloys of

known Pb activity. For antimony, e.m.f. studies suggested

Sb^I^ for solutions of Sb in Sbl^ (23), which agrees with

results of vapor pressure measurements (24).

While some of these studies were being carried out, the

effects of various solvents on the solubility of metals in

their molten salts were receiving attention. An interpreta­

4

tion in terms of acid-base reactions developed by Corbett,.

Burkhard, and Druding (1) has proven valuable for under­

standing post-transition metal solubilities and stimulating

further research. This interpretation recognized that for a

suggested equilibrium in the cadmium-cadmium(II) chloride

system, Cd° + Cd^^, a decreased solubility on addi­

tion of basic chloride ions in the form of KCl could be due

2+ to stabilization of the more polarizing Cd ion. Corre­

sponding increased solubilities on addition of acidic sub­

stances such as CeClg and AlCl^ were then attributed to

reduction of the effective basicity of chloride ions already

in the melt by the more polarizing or complexing Ce^"*" and

Al^^. Such "acid stabilization" has been substantiated by

the isolation of solid subsalts such as Cd2(AlCl^)2 (1),

GaAlCl^ (25), and "BiAlCl^" (26).

In this investigation, a combination of acid stabiliza­

tion with e.m.f. measurements was developed for the further

identification of species formed by solution of metals in

their molten salts. Previous workers have used e.m.f.

determinations to examine the species involved in molten

systems (16, 21, 23, 27). In general, this involves plot­

ting e.m.f. versus a logarithmic function of the activities

5

of the chemical units involved in the electrode reactions,

according to the Nernst equation

E = E° - RT/nF In £ E_

<2^ «B

where E represents the e.m.f. for the reaction aA + bB cC

+ dD (where a, b, c, and d are the number of moles of sub­

stances A, B, C, and D), E° is the standard e.m.f. for the

reaction, n is the number of electrons exchanged in the

reaction, R is the gas constant, T the absolute temperature,

F the Faraday constant, and the activity of substance X.

For constant temperature the slope of such a graph is equal

to -RT/nF, and since R, T, and F are known, n may be deter­

mined.

At low concentrations, the activities in the Nernst

equation can often be replaced by concentrations without

serious effects on the result. For instance, Flengas and

Ingraham (28) have shown that the activity coefficient for

Ag"^ in NaCl-KCl (1:1) remains constant to at least 6 mole

percent, and deviations in e.m.f. studies on cobalt occur

only in excess of 4 mole percent.

A review of previous electrochemical studies in molten

salts showed that the reference couple Ag:AgCl ('-17o) in'

6

halide melts had been frequently and successfully used.

Junctions employed varied from the fine porosity frit used

by Hill, et al. (29), to the asbestos fiber sealed through

glass used by both Senderoff and Brenner (30) and Flengas

and Ingraham (28), to the thin glass bubble developed by

many workers and applied specifically to dilute Ag"*" solutions

by Bockris, et al. (31).

Laity (32) has discussed the junction potentials asso­

ciated with glass diaphragms. At the contact between the

diaphragm and the melt, very large concentration gradients

occur for the conducting species of both the melt and the

glass. An accurate description of what actually occurs in

such a case would be difficult to achieve. But, if identical

melts are placed on both sides of the diaphragm, the gradi­

ents at the second glass-melt contact are exactly opposed to

those of the first contact, and junction potentials can be

expected to cancel. This proposition is supported by mea­

surements on the two cells, Ag AgCl glassjAgCljCI2 and

AgjAgCl Clg (32), which produced the same results. Addi­

tional evidence was provided by Kolotii (33), who found no

disparity between asbestos and Pyrex or molybdenum glass

junctions used in cells containing molten PbCl2-NaCl.

7

For this study NaAlCl^ was chosen as a solvent which

could provide acid stabilization. In order to make sub­

stitution of concentrations for activities reasonable and to

eliminate junction potentials, elements of interest were to

be added to NaAlCl, as chlorides or tetrachloroaluminates in

amounts of only a few mole percent. The addition of 5 mole

percent AICI3 was planned to insure the acidity of the melt..

Even at the lowest temperature possible for the suggested

NaAlCl^-AlClg melt (34), its AlCl^ vapor pressure is of the

order of 1 to 8 mm. (35, 36). Therefore, any cell containing

the NaAlCl^-AlCl^ melt had to be sealed, or significant

losses of AlClg would have resulted in a short period of time.

Due to the low solubilities of some of the metals of

interest in their salts and the dilution with 95 or more

percent NaAlCl^-AlCl^, the problem of making proportionate,

small changes in concentration of reductant was extreme.

Since the addition of metals or salts in small enough amounts

would be clearly impractical, some method such as coulometric

production was essential.

The cell upon which measurements were made was thus of

the following type:

4

in NaAlCl/, I

8

where M is the metal of interest and I is an inert electrode.

Some aspects of the approach outlined above have been

successfully used for studies of lower oxidation states in

molten halides by Topol and co-workers (16, 27). Their

measurements were made in three-compartment cells which

contained asbestos fibers or ultrafine frits for junctions.

The reference was usually either a .5 mole percent or a

saturated solution of the metal of interest in its molten •

halide. Concentrations were changed coulometrically, and

the cells were operated at temperatures between 240 and 700°,

generally under a flow of argon.

The first elements chosen for this study were cadmium

and lead. Based on past studies of cadmium in the closely

related Cd(AlCl^)2 melt (9), it seemed likely that the

species formed by interaction of cadmium and cadmium(II) in

NaAlCl^ would also be Cd^"*". A high solubility (~40 mole

percent dissolved metal) of cadmium relative to cadmium(Il)

was also probable in the NaAlCl^ solvent. These factors

meant that equipment and procedure could be developed with

a minimum of problems while the cadmium species was being

studied under different solvent conditions. Evidence for a

reduced lead species had also been found, but for the less-

9

closely related pure chloride melt (16, 22). Also, the

solubility of lead in its chloride (.006 mole percent at

500° (37)) is much less than that for cadmium in its

chloride, so the effectiveness of the combination of acid

stabilization and e.m.f. measurements could be further

tested.

Subsequent experiments were carried out with tin,

nickel, and zinc. Tin was selected because of its group

relation to lead and because of a lack of published informa­

tion about the tin reaction. Nickel was chosen as the

first-transition-series element most likely to form measure-

able amounts of reduced species. This choice was based on

the high solubility of nickel in its molten chloride (9.1

mole percent at 978° (38)) and on Born-Haber cycle calcula­

tions of the stability of first-transition-series elements

in various oxidation states (39). Zinc shows a very low

solubility in its chloride (.18 mole percent at 500° (37)),

but its group relation to cadmium and mercury is of interest

because the latter elements both form the dimerized reduced

2+ species, M2 .

10

EXPERIMENTAL DETAILS

General Cell Design

The cell used in this study (Figure 1) consisted of

three compartments constructed primarily of Pyrex glass.

The indicator electrode compartment was flanked on one side

by the reference electrode compartment and on the other side

by the working anode electrode compartment. The junction

between the reference and indicator compartments was a thin

glass diaphragm, and that between the indicator and working

anode compartments was a 10 mm. diameter, ultrafine,

sintered-glass frit. The coulometric reduction in these

experiments was carried out against the working anode. This

procedure allowed the reference half-cell to remain undis­

turbed during concentration changes in the indicator compart­

ment .

Wires leading to the electrodes within the cell were

silver. These entered the cell through 1/8 inch Kovar

metal-to-glass graded seals soldered around the wire. About '

15 cm. of each wire formed a coil inside the cell, and the

end of the coil was attached to the electrode. The elec­

trodes nearly touched the bottom of the cell and were between

11

Figure 1. General cell design; a - reference compartment, b - indicator compartment, c - working anode compartment

12

2 1/2 and 4 cm. long.

A sidearm of 10 mm. glass tubing was connected to each

cell compartment at the base of the cell, perpendicular to

the plane of the three electrodes. These sidearms served as

a place in which to freeze the melts at the completion of

the experiment in order to protect the fragile diaphragm. A

4 to 6 mm. glass tubing connected the indicator and working

anode compartments 2 1/2 to 4 cm. above the bottom of the

cell and served as a pressure equalizing passage between the

two compartments.

Ring-sealed to the outside of the cell just below the

Kovar metal-to-glass graded.^ seal was a 20 to 24 cm. length of

10 mm. glass tubing which shielded the silver wire from the

corrosive, conductive liquid in the thermostat.

Cell Components and Construction

The diaphragms were made by heating the closed end of a

10 mm. diameter tube and sucking on the blow tube to form a

thin bubble inside the tube (Figure 2, a). Another tube was

later butt-sealed to the end (b). The resistance of each

diaphragm was tested in molten NaNO^ containing several per­

cent AgNOg, using silver wires and a vacuum-tube voltmeter.

13

10 mm

N 10 mm

13mm

Figure 2. Details of cell construction: a - diaphragm formed in end of glass tube, b - tube butt-sealed near diaphragm, c - Kovar metal, d -graded seal, e - shield tube joint, f - shield tube, g - coil of silver wire, h - vacuum tight soldered seal, i - addition of electrode to cell

14

For these measurements, special precaution was required to

remove air trapped inside the bubble. Also, a pipette was

used to fill the inside of the tube so the entire tube did

not need to be submerged, which would have allowed conduc­

tivity around, rather than through, the diaphragm.

Temperature dependences of a variety of diaphragms were

plotted so that the behavior of others could be roughly

estimated from a single resistance measurement. Satisfactory

diaphragms had no visible imperfections and had resistances

of 60 Kn or less at the temperature at which they were to be

used. Such diaphragms allowed cell voltages to be determined

accurately to the nearest millivolt or better. The dia­

phragms were washed extensively with water to prevent a

yellow coloration of the glass by silver during later glass-

blowing.

The soldered seals where silver wires passed through

the Kovar metal-to-glass graded seals (Figure 2, c and d)

were also pre-tested. The Pyrex end of the graded seal was

joined to a glass tube 13 mm. in diameter, and the shield

tube, 10 mm. in diameter, was sealed at the same joint (e).

The long end of the shield tube was then cracked off at (f)

to expose the Kovar metal tube. About 15 cm. of a 45 cm.

15

long silver wire (99.9% .025 inch diameter, from American

Platinum Works) was wound into a coil (g) which would fit

inside the 13 mm. tube. Then the wire was threaded through

the Kovar tube, the tube pinched against the wire (h) and

soldered closed (#655 Ag-Cu-Mh-Ni braze, m,p. 752°, from

Handy and Harmon). The lower tube was connected to a vacuum

system, and the soldered seal was tested for leaks. If it

remained below discharge when held overnight at 300°, it was

used in a cell after the end of the shield tube was recon­

nected at (f).,

Next, the parts of the cell other than electrodes were

carefully combined while the 15 cm. silver wire remained

coiled and thus out of the way of glassblowing. The small

tube, which connected the indicator and working anode com­

partments (Figure 1) was added to the cell to equalize AlCl^

vapor pressures over the melts in the two compartments.

Previously, small inequalities in the percentage of AlClg

added to the two compartments had forced the melts through

the frit in one direction or the other.

A sample of each electrode material was tested for

inertness in the NaAlCl^-AlCl2-MCl2 melt, where M represents

the metal of interest. Zinc was rejected because it reduced

16

aluminum from the melt. Other materials showed no weight

loss on 24 or more hour contact with the melt at 300°.

Materials that form intermetallic compounds or large amounts

of solid solution with the metal of interest were not used

in the indicator compartment since such interactions would

have occurred during coulometric reduction. For situations

not covered by Hansen (40), weight loss tests were run in

NaAlCl^-AlCl2-MCl2 melts at temperatures above the melting

point of the metal M.

On these bases, tantalum was used as the indicator

electrode for measurements on cadmium, lead, and zinc;

tungsten or carbon was used for tin studies. Tantalum is

known to form an intermetallic compound with aluminum (40),

but no apparent problems related to this compound arose.

Also, the tantalum electrodes themselves appeared unaffected

by cell operation. If the rate of coulometric reduction

during cell operation had been higher, concentration polar­

ization of the reacting ion might have occurred, and the more,

electropositive aluminum would have been reduced.

The electrode for the working anode compartment was

commonly composed of the same metal as that being studied.

Early in the work it seemed desirable to use a metal here

17

which was more active than the metal being studied, as

indicated by Delimarskii's electromotive series in NaAlCl^

(41). Such an anode produced ions incapable of interaction

with the reduced species coulometrically produced in the

indicator compartment. However, attempts using aluminum

and thallium produced unexplained problems, suspected to be

due to diffusion of unknown active species through the frit.

In order to eliminate these problems, two changes were

made to the working anode compartment. The working anode

electrode was thereafter composed of the same metal as that

being studied. Secondly, the salt of the metal of interest

was placed in the working anode compartment at the same

concentration as that placed in the indicator compartment.

In order to restrict diffusion further, the pores of the

frit were partially heat-collapsed for several cells. The

frit resistances produced in this manner showed extreme

variations, so subsequently, unaltered frits were again used.

The cadmium and thallium electrodes were 3 mm. in

diameter and were cast from 99.999 and 99.9% metal respec­

tively. The tantalum electrodes were made of 1/8 inch

diameter tantalum rod, the tungsten electrodes of 1/16 inch

diameter Linde Heliarc, pure tungsten electrodes with ground

18

finish, the silver electrodes of 10 gauge wire, the aluminum

electrodes of 99.99%, iron-free, 1/8 inch diameter wire

(A. Do Mackay), and the carbon electrodes from spectroscopic

grade carbon rod.

Each electrode was joined to a few centimeters of silver

wire as follows: the silver, lead and cadmium electrodes

were melted at one end, and the wire was pushed into the

electrode a short distance; a small groove was filed around

the tantalum electrode, and the silver wire was wrapped

tightly in the groove and joined to itself by melting; nickel

was plated on the end of the tungsten electrode, and the

silver wire was joined to the nickel by melting; the wire

was pushed into a tightly fitting hole in the end of the

carbon electrode.

Finally, a small hole was blown at the bottom of each

compartment (Figure 2, i), and the end of the Ag wire coil

was pulled out through the hole. The short wire attached

to the electrode was joined to the coil wire by melting.

Each electrode and wire was pushed back into the cell,

collapsing the Ag coil. Electrical continuity between the

electrodes and the wires outside the cell was checked. The

small holes were then sealed.

19

Experimental Procedure

The cell was thoroughly washed with dilute acid and

water and then quickly dried by evacuation so as to minimize

oxidation of the Kovar and other metals involved. Both sides

of the diaphragm were evacuated or returned to atmospheric

pressure simultaneously, since the diaphragm was often too

fragile to survive a large pressure differential.

Chemical materials were loaded into the cell through

the sidearms. The transfers were made in an argon-filled

dry box fitted with an evacuable entrance lock, a circulating

system using a blower and Linde 4A and 13X molecular sieves,

and an open flat tray containing A Welch triple-beam

balance (calibrated to .01 grams) within the dry box was used

for all weighings.

Each compartment was loaded with 4 to 6 grams of NaAlCl^

and .2 to ,3 grams of AlCl^. On the order of .3 grams of

AgAlCl^ were added to the reference. Equal amounts of the

chloride or tetrachloroaluminate of the metal being studied

were added to the indicator and working anode compartments.

The cell was evacuated and subjected to dynamic high vacuum

(below discharge) for several hours. Then the sidearms were

20

sealed about 10 cm, from the electrodes.

The silver wires of the cell were joined (by melting)

to the silver wires leading to the potentiometer. To

minimize thermal shock, the cell was clamped directly above

the thermostat for several minutes, then slowly lowered into

the bath. At first the cell was tilted so the melts re­

mained in the sidearms and did not contact the electrodes.

If leaks existed where the sidearms had been sealed, bubbling

occurred in the melts at this time. Also, the presence or

absence of any slowly-dissolving materials was observed. The

cell was then tilted so the melts surrounded the electrodes,

taking care that the inside of the diaphragm was filled with

melt rather than vapor.

At this point the resistances of the diaphragm and frit

were measured with a vacuum-tube voltmeter. Because of the

potential of the cell itself, only a rough estimate could be

obtained by reversing the voltmeter leads and averaging the

results. The diaphragm resistance was within 10 KQ of that

measured during cell construction. The resistance across the

cell wall to the thermostat melt was two to three times the

resistance across the diaphragm. The resistance of the frit

was usually of the order of 10 to 100 ohms. For a partially

21

heat-collapsed frit, the resistance calculated using Ohm's

Law and the voltage and current measured during electrolysis

was 31 Kn versus 28 KQ measured directly during e.m.f.

measurements.

Voltages were followed until equilibrium was reached,

as indicated by constant voltage. This usually required

about 24 hours. Concentrations were then changed coulomet-

rically at a rate of ,05 microequivalents per second for ,1

to 10 microequivalents. During the measurement period the

cell was "rocked" several times an hour to aid in reaching

constant voltage.

When the working anode half-cell was altered to reduce

suspected diffusion, a saturated electrode of the metal being

studied was thereby established in the working anode compart­

ment. If the half-cell voltage of this saturated electrode

were not affected during small amounts of coulometric con­

centration changes, it could also serve as a reference for the

indicator half-cell. Therefore, in the remaining experi­

ments, the e.m.f.'s of both the indicator-reference and

indicator-working anode couples were monitored.

22

Materials

All materials were purified in glass using common vacuum

line techniques and electric resistance furnaces connected to

controllers and/or Powerstats. In order to eliminate im­

purities, the glass sublimation or distillation tubes were

washed and then heated with a torch or baked in a furnace

while under a dynamic vacuum which was held below discharge.

Materials were stored in the original, evacuated preparation

tubes or in argon-filled sample containers. All transfers

were made in the dry box.

Aluminum trichloride

Reagent grade AlCl^ was doubly sublimed under 10 to 15

cm. pressure of helium at approximately 180°« The first

sublimate was allowed to form a thick deposit and was then

sealed under vacuum in the glass tube. The ampoule thus

formed was placed in a second sublimation tube where, upon

heating, the thick deposit fractured the ampoule. The

temperature of fracture varied widely, occurring at 120 to

155°. Rate of heating was no doubt a factor in this varia­

tion, since Foster (42) has reported that AlCl^ is a poor

heat conductor. The residue of the second sublimation was

23

merely a thin shell of oxide. This procedure produced a

white product which contained no traces of yellow impurity.

Sodium tetrachloroaluminate

A weighed amount of reagent grade NaCl was placed in one

end of a glass tube which contained a medium-grade frit. The

salt was then dried overnight at 480° under dynamic vacuum,

A stoichiometric quantity of AlClg was added to the same tube,

and the tube was then evacuated'and sealed. This mixture was

I

heated above the melting point of AlCl^ and shaken until the

NaCl dissolved. The melt at this point was commonly brownish

and could not be cleared by addition of either aluminum or

sodium. Therefore, according to a procedure suggested by

Morrey (43), the melt was digested at 450° for one or more

days and filtered to remove a black solid which agglomerated

during digestion. This solid was insoluble in common acids

and volatilized without residue upon heating, indicating

that it was probably carbon. The NaAlCl^ product was a white

solid which produced a water-white melt.

CadmiumC 11") chloride

Reagent grade CdCl2 was doubly sublimed under vacuum at

500° in an 18 mm, diameter glass tube which was partitioned

in two places by small (7 mm, diameter) tubing, so the

24

residue could be sealed and discarded after each sublimation,

A supplementary furnace was used to maintain the empty part

of the tube at 300°. This prevented a small amount of

cadmium metal, which commonly sublimed at the beginning of

the operation, from depositing where CdCl2 would later

deposit.

CadmiumC11) tetrachloroaluminate

A stoichiometric mixture of AICI3 and CdCl2 was sealed

in glass, heated above 200°, and rocked until all the CdCl2

dissolved. This produced Cd(AlCl^)2, previously reported by

Corbett, Burkhard, and Druding (1). It was later found

desirable to digest and filter the product to remove carbon,

but undigested salt was used in this study.

LeadCII) chloride

PbCl2, obtained as a Pb-PbCl2 mixture from Corbett and

von Winbush (12), was distilled at 920° under 75 mm. pressure

of chlorine. After transfer to a new distillation tube, the

product was again distilled under chlorine at 920°, A trans­

parent white solid resulted,

Lead(Il) tetrachloroaluminate

Stoichiometric amounts of PbCl2 and AlCl^ were heated

together. When the AlCl^ melted at 193°, it reacted to form

25

a white solid which did not melt though the temperature was

raised to 260°. In order to produce a homogeneous melt

without encountering the dangerous pressures of AICI3 which

occur above 250° (35, 36), the tube was pushed gradually into

a furnace held at 340°. The product was digested at 450° and

filtered through a medium-grade frit to produce a white

solid. The melting point was found to be congruent at 275 +

2°. The product was also distinguished from PbCl2, AlClg,

and physical mixtures of PbCl2 and AlClg by powder patterns.

On one occasion a small quantity of an unknown liquid was

formed by the procedure outlined here. Digestion of

Cd(AlCl4)2 preparations also frequently produced this

liquid. Since the liquid had a high vapor pressure at room

temperature and a freezing point considerably below 0°, it

was evacuated from the Pb(AlCl^)2.

Pb(AlCl^)2 has been reported previously by Jander and

Swart (44), who obtained it as the product of a conducti-

metric titration in SbClg. Their compound, however, was

^Barnes, R. D., Ames Laboratory, U. S. Atomic Energy Commission, Iowa State University of Science and Technology, Ames, Iowa. Digestion of Cd(AlCl4)2 preparations. Private communication. 1964.

26

black and thus most likely contaminated.

Tin(11) chloride

SnCl2 5 obtained as a Sn-SnCl2 mixture from Corbett and

von Winbush (12), was distilled at 920° under 75 mm. pressure

of chlorine. After transfer to a new distillation tube, the

product was distilled a second time under chlorine at 920°.

A translucent, colorless solid resulted.

Tin(II) tetrachloroaluminate

Two moles of AICI3 and one mole of SnCl2 were heated to­

gether. The SnCl2 dissolved quickly when the AlCl^ became

molten. Overnight digestion at 390° and filtration through

a medium-grade frit produced a slightly greenish-colored melt

which froze to an off-white solid. Several preparations pro­

duced the same result. The compound melted congruently at

223 + 2° and was differentiated from SnCl2, AICI3, and

physical mixtures of SnCl2 and AlClg by powder patterns.

Zinc(II) chloride

Reagent grade ZnCl2 was distilled under vacuum at 350°,

transferred in the dry box to a new tube, chlorinated over­

night under 58 mm. of chlorine at 500°, and distilled again

at 420°. The product melt was clear and without solid

impurities.

27

NlckeKII) chloride

NiCl2, prepared by F. C. Albers, was sublimed once at

670° in a fused silica tube.

Silver(I) tetrachloroaluminate

Stoichiometric amounts of AlClg and reagent grade AgCl

were heated to 210° and rocked to dissolve the AgCl. The

product was then digested at 300° for several days and

filtered through a medium-grade frit. A clear melt and a

white solid resulted. Its melting point was congruent at

170 + 2°. Digestion at temperatures above 300° was avoided

because a yellow coloration of the glass container occurred

which indicated reaction with the container.

Other Equipment

In order to maintain constant temperature, the cell was

immersed in a molten salt bath while measurements were being

made. The bath was contained in a common enameled pot and

heated by a 1000 watt Edwin Wiegand heater, which was in

turn controlled by a Powerstat. The heater and bath were

insulated with firebrick, asbestos, and Transite. Bath

temperature could be controlled within + 1° by the Powerstat

and the amount of insulation used on top of the bath.

28

The bath consisted of a mixture of 73 mole percent

NaN02 and 27 mole percent KNO3. It was stable for over 24

months and suitable for temperatures as low as 195°, where

surface freezing began. However, the corrosive, conductive

nature of the melt necessitated using shields around wires

leading into the cell.

An arrangement of 1/2 inch aluminum rods allowed the

cell to be rocked by hand. The cell itself was held by a

Castaloy clamp connected to a long rod which pivoted over a

third rod as a see-saw.

Voltages were measured Kith a Leeds and Northrup #7552

Potentiometer in conjunction with a Leeds and Northrup #2430

Galvanometer. These instruments were necessary because of

the high resistance of the diaphragm. The power source was a

Willard 2 volt lead storage battery; the standard cell was

made by Eppley Laboratory. No. 26 copper wires joined the

potentiometer and the silver wires leading to the cell.

The copper-silver junctions were outside the heated region

of the bath, and therefore thermocouple e.m.f.'s were

avoided. A reversing switch was placed in the copper section

of the wires to facilitate measuring voltages of different

polarities. A second switch allowed measurement of either

29

the indicator-reference or the indicator-working anode

voltage. All wires were protected by insulation sleeves.

Coulometric reductions were made with a Model IV Sargent

Coulometric Current Source capable of delivering as little

as .05 (+ .17o) microequivalents per second for intervals as

small as .1 seconds.

30

RESULTS AND DISCUSSION

Data Treatment

The data were analyzed by a plot of log versus E,

where E is the voltage of either the indicator-reference

couple or the indicator-working anode couple, and is the

total amount of reduction carried out in the indicator com­

partment up to that time. This relation may be derived in

the following way.

For any reducible, divalent ion present in the indicator

compartment, coulometric reduction will produce an undesig­

nated, reduced ion according to the reaction

+ ne" - #(2-^) + ̂

where M represents the metal atom and n the number of elec­

trons (e~) involved in reduction of each To account for

possible catenation of the reduced species, the reduction can

be rewritten as

m2+ + ne- - 1/m

or 1/n + e~ -«• 1/mn

31

where m is the degree of catenation. Thus, for every

equivalents of current passed through the cell, q^/mn equiva­

lents of the reduced species will be formed, as

Therefore, if the original concentration of divalent ion is

Cq, the concentration will be (C^ - q^/n) after qj_ equiva­

lents of reduction. Similarly, if the original concentra­

tion of reduced species is zero, its concentration after q^

equivalents of reduction will be q^^/mn equivalents.

The potential observed for the indicator-reference

couple can be written as

associated with the reference are assumed to be constant.

Separating all the constant terms as C gives

E = E° -2.3 RT/nF

for the reaction 1/m + nAg+ + nAg°. The

terms m and n, the junction potential Ej, and the activities

E = -2.3 RT/nF log + C »

m

32

Assuming that, as noted in the Introduction, activities may

be replaced by concentrations for the low concentrations

used, and substituting the terms derived previously for these

concentrations, give

E = -2.3 RT/nF log + C,

[q^/mn]^'^™

or E = 2,3 RT/mnF [log q^ - m log [C^ - q^/n] - log mn] + C,

Remembering that mn can be treated as a constant,

E = 2.3 RT/mnF [log q^ - m log [C^ - q^/n]] + c ' . (1)

The second log term is also constant for values of q^^ which

are very small compared with Cq, so

E = 2.3 RT/mnF [log q^] + c". (2)

Thus, a graph of log q^ versus E will yield a slope of 2.3

RT/mnF. All factors of this last expression except m and n

are known, so the mn product may be found.

If q^ is not small compared with C^, a plot of log q^

versus E will produce a curve, rather than a straight line.

Therefore, the data must be plotted according to Equation (1)

which takes into account the effect of q^ on . This

33

procedure, of course, demands an assumption of m and n

values. The catenation of the reduced species, m, is rea­

sonably limited to small whole number values, while n may-

vary between zero and two. Further, the product mn must be

integral. These considerations are necessary for the treat­

ment of the cadmium data.

If the working anode remains continuously saturated and

diffusion and solute migration are not important, it has a

constant potential analagous to the reference, and a graph of

log versus E for the indicator-working anode couple also

yields mn.

Sodium ions were assumed to provide most of the ionic

exchange through the frit required to retain charge equality

in the indicator and working anode compartments during

coulometric reductions. This dilution was considered negli­

gible, since it amounted to only .1 mole percent of the

indicator compartment melt for a typical experiment.

In the case that more than one reduced species exists

in the melt in the concentration range studied, it is pos­

sible that two or three linear portions of the plot might

appear, each corresponding to an mn value for a major species

at concentrations where it prevailed. If comparable amounts

34

of two reduced species have been formed, such experiments

would not be able to distinguish between them.

Initial Impurities

In every case the first few points showed a negative

deviation from the linear portion of the log qj_ versus E

plot. It is unlikely that a species in a different, lower

oxidation state was the cause of these deviations, since the

slope would correspond to a higher mn value and therefore

correspond to a more complex species which became simpler at

high concentrations. Also, the deviated points formed a

curve, rather than a straight line, and the deviations were

much the same for all experiments.

Topol (27) was able to polarographically detect a

reduced bismuth ion during an e.m.f. study of the Bi-BiClg

system that had similar negative deviations. In subsequent

e.m.f. studies in which an initial impurity was also

apparently present, the amount of this initial "impurity"

of the reduced species was calculated from the first datum

and a presumed mn data (16). A correction factor was then

added which brought the first few points into line with the

remainder of the data. In these later experiments, however,

no attempt was made to show that an oxidizable entity was

indeed present.

In this study the initial voltage observed upon immer­

sion of each cell in the thermostat drifted in a positive

direction. This change is in the same direction as that

caused by coulometric production of a reduced species.

Equilibration of the melts in the sidearms during this

initial period did not significantly affect the amount of

the drift, so diffusion through the frit and interactions

with the electrode materials could not have been responsible.

Thus, it is likely that a reduced species was being produced

during the initial decay.

Further experiments would be required to decide whether

reducing impurities present in the components or some inter­

actions with the glass container were responsible. In view

of difficulties encountered in purification of AlCl- and

the subsequent appearance of a black solid in all digested

tetrachloroaluminates made from AICI3, this material is

suspect. It is possible, as well, that the e.m.f. drift

and the initial deviations of the plot were unrelated. In

this case, perhaps a different electrochemical reaction was

responsible for the early deviations but was later largely

36

drowned out by the reaction of interest.

Corrections for the effects of initial impurities were

calculated. The correction factor for this impurity must be

added to all qj_ values, not just the first. This means that

the slope of the uncorrected graph in each case is higher

than the correct slope, since the plot involves a log func­

tion. The integral mn product obtained from the log qj_

versus E plot was assumed to be correct. Then (q^ + qo)s

where q^ was the initial impurity, was substituted for qj_

in Equation (2), and q^ was calculated using E and qj_ values

from the first and last points on the graph. For cadmium,

Equation (1) was employed in a similar manner.

The calculated initial impurities so deduced were much

smaller for the lead and tin experiments than for the cad­

mium determinations. The average value for cadmium was .02

mole percent of the melt, while that for tin and lead was

only .007 mole percent. The latter value is considerably

better than the .01 to .02 mole percent found by Topol (16).

9 4" Considered as a percentage of C^, the initial M concentra­

tion, cadmium impurities averaged 4.6 percent, while those

for tin and lead averaged only .14 percent.

37

Cadmium

Since cadmium has a high solubility in the melt used in

this study (43 mole percent dissolved metal, based on

Cd(II)), smaller initial concentrations of Cd^"*" (C^) were

used than in the lead or tin cells. This was an attempt to

minimize diffusion through the frit by. decreasing the con­

centration gradients across the frit. Therefore, for all

three successful cadmium cells, qj_ was not negligible com­

pared with Cq.

For Run 20, however, was only 4,6 percent of at

the final point, so only small errors were introduced by

considering that [Cq - q^/n]^ was constant. For the uncor­

rected indicator-reference data, a plot of log versus E

yields mn = 2.56. A correction of 7 t_i eq. was calculated

(see Data Treatment) assuming mn = 2 to be the correct value.

This is a reasonable choice since mn must be integral, and

even the 1.6 n eq. correction calculated using mn = 3 results

in a corrected mn nearer 2 than 3 (2.42). Further, a cor­

rection of 1.6 |_i eq. does not bring the early deviations into

line with the remainder of the data. The corrected data

(using 7 la eq.) yields mn = 1.98.

38

Further support for mn = 2 was obtained by using Equa­

tion (1) to analyze the data from Runs 8, 20, and 21, This

procedure requires a choice of m and n values, as outlined

previously (see Data Treatment). The species Cd°, Cd^"*",

4+ Cdg , etc. are all consistent with an mn product of 2,

Though the graphs for m = 1, n = 2 (Cd°) were slightly

curved, rough mn values between ~1,5 and ̂ 2.5 were calculated

from the various slopes possible. For instance, mn is 1.83

for points in the middle of the plot and mn = 2.35 for the

last few points for the indicator-reference data of Run 21.

The selection of m = 2, n = 1 (Cd^"*") produced straight

I lines which yielded sensible values for the mn product.

2 Figures 3 and 4 show typical graphs of log [Cq - q^] /q^^

versus E for data from both the indicator-reference and

indicator-working anode couples. The corrections for q^

have been applied as outlined previously, assuming m = 2,

n = 1. Table 1 gives the results for this choice of m and n.

Table 1 shows the agreement between the corrected mn

values for the indicator-reference couples of Runs 8, 20, and

21. The result from Run 8 is somewhat low, but the average

deviation is twice that for the indicator-reference results

of the other experiments. An average deviation of .002 volts

1.0

1.5

-Z2.0 cr

CVJ

T 2.5 o O

g 3.0

3.5 —

4.0

CADMIUM 275°

CORRECTED mn = 2.02

UNCORRECTED m n = 2.78

1

0.28 0.29 0.30 0.31 0.32 0.33 0.34 0.35 0.36 0.37 0.38

E(VOLTS)

Figure 3. Indicator-reference potentials for cadmium, Run 21

1.0

1.5

2.0

O O

O 3.0 o

3.5

4.0

1 r CADMIUM 275°

CORRECTED mn = 2.00

UNCORRECTED mn = 2.52

J 1 I L

o

0.14 -0.13 -0.12 -0.11 •0.10 -0.09 -0.08 -0.07 -0.06 -0.05 -0.04

E (VOLTS)

Figure 4. Indicator-working anode potentials for cadmium, Run 21

Table 1. Results for cadmium

Indicator-reference Indicator-working couple anode couple

Run 8 Run 20 Run 21 Run 20 Run 21

Uncorrected mn 1.93 2.73 2.78 1.50 2.52

Corrected mn 1.84 2.03 2.02 1.44 2.00

Average deviation from corrected plot (volts) .002 .001 .001 .006 .001

Cq (initial Cd^"*" concn.) (n eq.) 27 890 67 890 67

(mole %) .1 3.8 .2 3.8 .2

Calculated impurity (u eq.) .7 8 7.5 1.3 5.8

mole 7o of melt .0003 .03 .02 .01 .02

% of Co 2.4 .9 11 .1 8.6

Mole 7o dissolved Cd for final data point 18.5 2.3 17 2.3 17

Mole 7o dissolved Cd at saturation, based on Cd(Il) 14 40

Temperature 293 276 275 276 275

42

corresponds to an average deviation in mn of .1, which could

account for a large part of the difference between 1.84 and

2 . 0 0 .

The indicator-working anode result of 1.44 for mn from

Run 20 deviates markedly from the other runs, but this

result also shows by far the least linearity, as indicated

by the large average deviation. Also, the disagreement

between the calculated impurities for the two couples of Run

20 contrasts with the agreement for Run 21 and is further

evidence for undetermined problems in the use of the working

anode as a reference. The working anode compartment of Run

8 contained no Cd^"*" and an aluminum, rather than cadmium,

electrode. No data for the indicator-working anode couple

were taken.

A third species which is consistent with an mn product

4+ of 2 is Cdg , for which m = 3, n = 2/3. This choice of m and

n led to plots which had larger scatter than m = 2, n = 1

graphs. The mn product calculated from the rough slope was

~4 for Run 21 and ^^3 for Run 20.

A single plot of m = 4, n = 1/2, for Run 21 produced

mn = 6.4. Plots for m > 4 are expected to show even greater

inconsistencies between the mn corresponding to the slope

43

of the plot and the tnn = 2 values upon which the plot is

based.

The limit of maximum reduction was not reached for any

run. A saturation limit of 50 mole percent dissolved metal

2+ would correspond to reduction of all Cd to a monovalent

state. The saturation limits given in Table 1 were calcu­

lated from the value of log (corrected) which corresponds

to 0 volts for the indicator-working anode couple. The Run

21 limit of 40 mole percent dissolved metal (based on Cd(Il))

is comparable to the 43 + 3% determined by the loss in weight

of metal buttons equilibrated with the melt. The calculated

value for Run 20 is probably low because of the low value for

the slope upon which the calculation is based. Thus, the

indicator-working anode data are less consistent than the

indicator-reference data. However, the excellent agreement

between the saturation limit measured by loss in weight and

the saturation limit calculated from the indicator-working

anode data of Run 21, which also produced an mn value con­

sistent with that from the indicator-reference data, is

further confirmation of the conclusions of these experiments.

All three cadmium cells used tantalum indicator elec­

trodes, and cadmium was added to the cells as Cd(AlCl^)2»

44

In general, the calculated impurities in the cadmium studies

were far greater than those found with lead; results could

probably have been improved by the use of digested

Cd(AlCl4)2.

Studies in several different melts have now reached the

same conclusion regarding a lower oxidation state species of

cadmium. Only the Raman study by Corbett (9) has desig-

2+ n 2+ 4+ nated Cd2 unequivocally from the series Cd , Cd^ , Cd^ ,

etc., but data from CdClg (8) and CdCl^-KCl-NaCl (45), as

well as Cd(AlCl^)2 (9), are in agreement that a species from

this series is formed. Since this study in Cd(AlCl^)2-

NaAlCl^ (90 mole percent)-AlCl2 points to the same series

2+ 2+ and suggests Cd^ , it appears likely that Cd^ is indeed the

species commonly formed when Cd dissolves in melts contain-

ing Cd

Lead

The results for two lead cells are shown in Table 2. In

both cases q^ is negligible compared with C^, so the data

ere analyzed unambiguously by a log q^ versus E plot accord­

ing to Equation (2). Typical graphs for the indicator-

reference and indicator-working anode couples are shown in

w

45

Table 2. Results for lead

Indicator-reference Indicator-working couple anode couple

Run 14 Run 18 Run 14 Run 18

Uncorrected mn 1 .04 1 .26 1.24 1 .42

Corrected mn .90 1 .00 1.04 1 .08

Average deviation from corrected plot (volts) .003 .002 .003 .003

Cq (initial Pb^^ concn.)

(la eq.) 1760 1760 1760 1760

(mole 7o) 5 5 5 5

Calculated impurity

(ij eq.) .7 3 .2 .7 6 .5

mole 7o of melt .002 .009 .002 .02

% of Co .04 .2 .04 .4

Mole 7o dissolved Pb for final data point .2 .6 .2 . 6

Temperature 276 279 276 279

Figures 5 and 6. Impurities were calculated assuming mn = 1.

Any possible variations in the reference cell were

eliminated by using the same reference for both Run 14 and

LEAD 276" 0.8

CORRECTED mn=0.90 0.4

UNCORRECTED m n - 1.04 cr

-04

-Q8

-1.2 -0.09 -0.08 -0.07 -0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0 QOI

E (VOLTS)

Figure 5. Indicator-reference potentials for lead, Run 14

LEAD 276»

0.8 CORRECTED m n = 1.04

04

UNCORRECTED mn = 1.24 CT

O 3

-04 -

-0.8

_ 1,2 1 1 1 1 1 1 L I I

-0.36 -0.35 -0.34 -0.33 -0.32 -0.31 -0.30 -0.29 -0.28 -0.27 -0.26 E ( VOLTS)

Figure 6. Indicator-working anode potentials for lead, Run 14

48

Run'18. The indicator electrode .in both cases was tantalum.

Run 18 contained a new frit and an unused lead working anode.

Lead was added to the cells as Pb(AlCl^)2.

Attempted calculations for the limit of reduction

yielded meaningless values in excess of the initial amount

of Pb^^, but extreme extrapolations of the data were required

for these calculations. As for cadmium, the reduction limit

was not actually reached during cell operation.

The results clearly indicate that mn = 1 for the solu­

tion of lead in Pb(AlCl^)2-NaAlCl^ (90 mole percent)-AlCl3.

3-f 5_>_ The species Pb , Pbg , Pbg^, etc., are all consistent with an

mn product of 1, and no further restriction is possible on

the basis of these experiments. The results are surprising

because they indicate that a different species is formed

in Pb(AlCl4)2-NaAlCl4-AlCl3 than in PbCl2- For the latter,

both Topol's e.m.f. study and Egan's polarographic measure­

ments suggest that Pbl"*" (Pb°, Pbg^, etc.) is formed. .

Since all of the species possible for mn = 1 would be

paramagnetic while Pb^^ would be diamagnetic, qualitative

susceptibility measurements were made. These were carried

out on the related Pb(AlCl^)2 melt where the solubility of

lead (see Other Experiments) is higher than in the NaAlCl^-

49

diluted melt used for e.m.f. measurements. Unfortunately,

no difference could be detected between the saturated and .

pure melts at 300°. The sensitivity of the equipment was

checked with aqueous solutions containing specific concen­

trations of paramagnetic species, and these measurements

showed that the possible concentration of paramagnetic lead

species should not be detectable. However, the possibility

remains that more accurate susceptibility measurements would

provide confirmation for the existence of two different

reduced lead species in different melts.

Tin

A modified cell was developed and used for experiments

with tin because the data from three earlier cells, Runs 15,

16, and 19, were of dubious significance. The earlier cells

contained molten tin anodes and either tungsten or carbon

indicator electrodes. The molten tin was contained in a well

at the bottom of the working anode compartment. Electrical

contact between the silver entrance wire and the tin was

established by a tungsten rod shielded from the NaAlCl^-

AlCl3-Sn(AlCl^)2 melt by a glass tube collapsed around the

rod.

I

50

Graphs of the data from Run 16 were reasonably linear,

but mn for the indicator-reference couple was about 9 while

that for the indicator-working anode couple was 5.6. The

frit had been partially heat-collapsed to reduce diffusion

because results from Run 15 showed the indicator compartment

was initially saturated with tin. Consequently, the resist­

ance of the frit used in Run 16 was 28 KQ, and the potential

across the frit during coulometric reduction reached 150

volts compared to the 1 volt normally observed.

An uncollapsed frit was employed for Run 19. The

indicator-reference slope could be variously interpreted to

give mn = 4.5 to 7. The indicator-working anode plot was

curved so no mn could be determined.

Because of these results, the cell for Run 23 was

modified to essentially eliminate diffusion. The frit was

offset from the line joining the electrodes by placing it in

a U tube extending in the opposite direction from the side-

arms (see Figure 1). Qualitative determinations of the density

of the melt and the capacity of each cell compartment were

made. After loading with amounts of salts calculated to fill

the compartments to a designated level, the sidearms were

sealed at predetermined points. These adjustments allowed

51

the melts to be tipped into the sidearms with the electrodes

and diaphragm still immersed. In this position the frit was

not immersed, so no diffusion occurred during either the

initial decay or the necessary equilibrations after coulo-

metric reductions. Pressure equalization could also take

place through the frit so the pressure equalizing passage

was eliminated from the cell. During the relatively short

periods of time required for coulometric reductions, the

cell was tilted in the opposite direction to immerse the

frit, while the electrodes remained immersed. In both posi­

tions the electrodes were immersed to approximately the same

level.

Silver was used for the working anode and carbon for the

indicator electrode. Tin was added to the cell as SnCl^. In

order to minimize diffusion, no data were taken for the

indicator-working anode couple.

The results of Run 23 are shown in Table 3 and Figure

7. Corrections were calculated as outlined previously,

assuming mn = 3. This is reasonable since the corrected mn

is 3.26 even if m = 4 is assumed. The calculated impurity

was based on the second data point since the first point was

not consistent with the rest of the data and led to a cor-

52a

Table 3. Results for tin, Run 23

Uncorrected mn 3,55 Mole % dissolved Sn .45 for final data point

Corrected mn 2.92 Calculated impurity q

Average deviation from (|a eq.) 2»8 corrected plot (volts) .001 mole % of melt .004

CQ (initial Sn^"*" concn.) % of C^ .08 (M- eq.) 3477

(mole 7o) 5.4 Temperature 277°

rected plot which was not linear. This first point was

probably not significant because of the extremely low amount

of reduction (.25 eq.).

While not conclusive, the results for Run 23 suggest

that the lower oxidation state species of tin formed by dis­

solving tin in Sn(AlCl^)2-NaAlCl^ (90 mole percent)-AlClg is

one of the series Sn^, Sn^^, Sn^"*", etc. This is the first

evidence for a specific reduced tin species in a melt, so no

direct comparisons can be made with other work. Four

previous references (17 - 20) have postulated the existence

of a lower oxidation state of tin in melts in order to

account for various electrochemical observations. These

2.0

1.8

1.6

1.4

1.2

1.0

0.8

0.6

~ 0.4

g 0.2

0

-0.2

-0.4

-0.6

-0.8

TIN 277»

CORRECTED mm= 2.92

UNCORRECTED mm = 3.55

Ul N or

-0.04 -0.03 -0.02 -0.01 0 0.0! 0.02 0.03 0.04 0.05 0.06 E(VOLTS)

Figure 7. Indicator-reference potentials for tin, Run 23

53

include otherwise unexplained breaks in polarograms of molten

SnCl2 (17) and chronopotentiometric studies in the same melt

(18 ) .

As with lead, the e.m.f. measurements suggest species

which are paramagnetic. However, qualitative susceptibility

determinations similar to those carried out with lead were

also unsuccessful for pure and saturated Sn(AlCl^)2. Further

studies with more accurate equipment are in order.

This investigation has demonstrated the feasibility of

electrochemical measurements on reduced tin species, particu­

larly in the presence of added AICI3, which significantly

increases the concentration of the reduced species (see Other

Experiments). However, other such studies are still required

to confirm the somewhat unusual results found here.

Zinc

E.M.F. measurements were made on only one cell contain­

ing divalent zinc. A lead working anode was used because

zinc reduced aluminum from NaAlCl^. Zinc was introduced as

ZnCl2. No meaningful results were obtained because the cell

potentials drifted continuously. The direction of the drift

was consistent with diffusion of the reduced zinc species

54

through the frit.

In separated experiments ZnClg was found to be slowly

soluble in 93 mole percent NaAlCl^-7 mole percent AlCl^» The

limit of solubility was 4.2 mole percent at 300° and was very

temperature dependent.

Other Experiments

Attempts to study nickel failed because suitable con­

centrations of Ni^"*" in NaAlCl^-AlClg melts could not be

achieved. NiCl2 did not dissolve in AlClg to any appreciable

extent at 160°, 225°, or 240° when heated for three weeks.

Heating AlCl^ at 240° in one end of a tube and NiCl2 at 500°

in the other end also failed to produce significant reaction.

Two NiCl2-AlCl3 (1:2 mole ratio) mixtures were heated in

sealed tubes with 25 and 83 mole percentages NaAlCl^, re­

spectively. One end of each sealed tube was heated at 370°

and the other at 270°. This produced transparent yellow

hexagonal crystals, which powder patterns indicated were

merely recrystallized NiCl2. In many cases, a rose-to-salmon

coloration of the melt was produced, but the solubility of

NiCl2 remained very small. This is somewhat surprising since

the preparation and structure of Co(AlCl4)2 has been reported

55

(46). The solubility of NiCl2 in a melt composed of 93 mole

percent NaAlCl^-? mole percent AlClg was estimated at < 0.1

percent at 300°, in agreement with the report of Delimarskii

for pure NaAlCl^ (47).

In the course of this work, solubilities of the re­

spective metals in the pure tetrachloroaluminates and in 90

mole percent NaAlCl^-5 mole percent AlCl2"5 mole percent

tetrachloroaluminate were determined. The results, together

with other values from the literature, are contained in Table

4. All tetrachloroaluminates prepared were found to be

miscible with NaAlCl^.

The table shows that for tin and lead, as well as cad­

mium, the solubility is increased by adding AICI3 to the

metal chloride. This is further substantiation for the con­

cept of acid stabilization developed by Corbett, Burkhard,

and Druding (1).

If the same reduced species is formed in the pure tetra­

chloroaluminates as in the 90 mole percent NaAlCl^ melts, the

data indicate that percent reduction of the divalent cations

is increased upon dilution with NaAlCl^ for tin and lead, but

not for cadmium. If the cadmium dissolves according to the

reaction Cd° + Cd^"*" Cd2^, the equilibrium constant

56

Table 4. Solubilities in different solvents (mole percent dissolved metal)

M MCI. (Ref.) M(AlCl4)2 90 mole % NaAlCl^ (Ref.) 5 mole % AICI3

5 mole % M(AlCl4)2

Cd 15 @600° (1) 40 (3335° (1) 3.8^ + .3 or 43 + 3^ @300^

Pb .006 @500° (12) .2 @305° .16& or 2.4^ @300°

Sn .003 @500° (12) .1 @290° .06* or 1.1^ @305°

^Based on total melt.

"Based on moles of M(II) only.

K = ^ is (as observed) not affected by dilution. [Cd2+]

However, the simplest proposed species for lead and tin cor-

r pb^i^ respond to the equilibrium constants K = — and

[Pb^+l

K = —5 for which the mole percent dissolved metal

[Sn^-]

based on M(II) must increase (as observed) with dilution if

the K is to remain constant. The measured solubilities are

thus in accord with the designation of mn = 2 for cadmium and

mn = 1 and mn = 3 for lead and tin, respectively. Calcula­

tions show that the increased solubility caused by dilution

57

would be .9 mole percent dissolved metal (based on M(II)) in

the 90 percent NaAlCl^ melt for lead and „48 mole percent for

tin for the two proposed equilibria. Since the measured

solubilities are better than twice these amounts, other

factors involving the altered ionic environment of lead and

tin ions in the 90 percent NaAlCl^ melt compared with the

undiluted tetrachloroaluminates must also have an effect. In

the pure tetrachloroaluminates there are effectively two

AICI3 molecules competing with each for basic Cl" ions,

while in the 90 percent NaAlCl^ melt there are three AICI3

molecules per due to the addition of 5 mole percent

AICI3. This increase in acidity might reasonably have little

effect on the cadmium equilibrium which is already shifted

2+ far toward Cd2 , but yet account for the increased solu­

bility of lead and tin.

Proposed Research

Further e.m.f. studies with several of the post-

transition metals would be of interest. In particular,

complex, reduced species of indium and bismuth should be

detectable. Since phase studies involving In(I) in chloride

melts show four compounds other than InCl, it seems likely

58

that ions related to these compounds occur in the melts.

Bismuth systems have already been extensively investigated,

but ions such as Big* and Big"*" are interesting possibilities

for e.m.f. determination. For gallium, there is no reason to

expect other than the simple Ga"*" ion, but its confirmation

would be relatively easy. These elements are most likely

susceptible to e.m.f. studies with approximately the same

equipment and techniques developed in this work. The

modified cell used for tin should also eliminate or pin­

point the problems encountered with zinc.

For the remaining elements, direct measurements of the

solubility of metals in their salts should be made prior to

more e.m.f. studies. Measurements in this study show that

the concentration of dissolved lead and tin in their

chlorides can be increased significantly by the addition of

AICI3 or NaAlCl^ plus AICI3. Therefore, the effect of added

AlClg or NaAlCl^ on metal solubilities in such halides as

MhCl2, CrCl2, C0CI2, AgCl, ZnCl2, Hg2Cl2, and TlCl should be

*Corbett, J. D., Ames Laboratory, U. So Atomic Energy Commission. Iowa State University of Science and Technology, Ames, Iowa. The reduction of BifAlCl^)] with excess bismuth metal. Private communication. 1964,

59

determined. In this connection, zinc must be measured by

some method other than weight loss, since zinc metal reduces

aluminum from AICI4 melts. Germanium halides are probably

too molecular and Ge(II) too weak an oxidizing agent to show

reduction by germanium metal, but addition of AICI3 or

NaAlCl^ might be tried. In any case, reduction limits for

most post-transition elements are known to increase from

chlorides to iodides, so the effects of addition of All^ and

NaAlI^ to the iodides might be interesting. The effects of

mixing halides are also as yet unknown. The substitution of

other acids for AlCl^ has been explored to some extent (8),

but various silicate melts remain as possibilities.

Additional e.m.f. studies might be more successful than

otherwise if several factors are considered. With the

present cell, some balance exists between eliminating

junction potentials and increasing the concentration of the

oxidized ion to allow higher absolute quantities of reduced

species. For systems which have even lower solubilities than

those already studied, the amount of oxidized ion can perhaps

be increased above 5 mole percent without introducing errors.

Within limits of salt preparation and thermostat size, the

amount of melt used in a cell may also be increased.

60

In this connection, e.m.f. measurements might detect a

reduced aluminum species, if it could be formed in suitable

quantities. The calculated solubility of A1 in AlClg (^-lO"^

mole percent at 230° (11)) is extremely low. Also, aluminum

has been used previously as an electrode material in NaCl-

KCI-AICI3 melts by Verdieck and Yntema, who did not note any

special problems (48). However, in the present work, sur­

faces of aluminum anodes disintegrated in NaAlCl^. On one

particular occasion with a cell for measurements on lead,

aluminum crystals were formed on the working anode side of

the frit, and lead crystals on the indicator side. This

occurred before coulometric reductions were begun. Both of

these observations suggested that a reduced aluminum species

was formed in the working anode compartment. Studies would

no doubt be more successful if iodides could be used, because

the solubility of A1 in AII3 is .02% at 230° (37).

The designation of a reduced nickel species may also be

possible, since Verdieck and Yntema (48) reported 1 mole

percent solubility in a 66 mole percent AICI2-20 mole percent

NaCl-14 mole percent KCl melt at only 156°. Substitution of

NaCl-KCl-AlClg for NaCl-AlCl^ would not require any radical

cell changes.

61

The Ag:Ag^ half-cell has been shown in this work to

provide a more consistent reference than the saturated work­

ing anode compartments. However, the anode compartment can

profitably be used as a reference for preliminary measure­

ments in systems where solubilities are as large as those

found for lead. This will allow simpler cells to be used for

part of the work.

Higher purity materials are still desirable. The un­

known liquid which was present at room temperature in one

Pb(AlCl4)2 preparation and the initial "impurities" con­

sistently observed both indicate that purification of

materials should continue to be an important part of further

e.m.f. work.

A continuously reading or recording electrometer would

greatly facilitate making measurements but probably cannot

be justified because of expense.

Certainly the possible tetrachloroaluminate compounds

have yet to be exhausted. Known phase diagrams for metal

halide-metal tetrachloroaluminate or metal halide-aluminum

chloride systems are sparse, and though many such diagrams

are no doubt simple, others might be interesting.

Further confirmation of the results of this work can be

62

obtained by polarographic measurements using approximately

the same cell. There is some evidence that decomposition

potentials can be determined through a glass diaphragm if the

melt contains sodium ions (49). Otherwise, an asbestos fiber

sealed through glass, such as that used by Topol and Oster-

young (50), can be substituted for the diaphragm. In addi­

tion to the large electrode used for coulometric reductions,

a solid microelectrode would have to be added to the indi­

cator compartment.

Vapor pressures, freezing points, or other properties

of the solvent for even the undiluted tetrachloroaluminates

would not be affected enough to infer particular reduced lead

or tin species.

Since this study indicates that paramagnetic species are

present in the solutions of lead or tin in their tetrachloro­

aluminates plus NaAlCl^, careful susceptibility measurements

with good equipment could provide interesting results.

63

SUMMARY

EoM.F. measurements were used to designate the lower

oxidation state species formed when cadmium, lead, and tin

dissolve in NaAlCl^ melts containing divalent tetrachloro-

aluminates of those metals. The effectiveness of the addi­

tion of the acid AICI3 to increase the solubility of metals

in their molten chlorides was further substantiated. Simi­

larly, the usefulness of acid stabilization to increase the

concentration of lower oxidation states for electrochemical

measurements was demonstrated.

Sealed cells of the type

Agi Ag""" (5 mole %) in NaAlCl^ glass M^"*". (5 mole %) ,

n)+ NaAlCl 4 Ta or C

were employed for M = cadmium, lead, tin, and zinc at 275 to

305°. The acidic solvent NaAlCl^ was chosen to stabilize the

lower state species, The concentration of this

reduced species was increased coulometrically against a third

half-cell connected to the cell by an ultrafine frit. Thin

glass bubbles were used for the reference junctions.

The data were analyzed by the use of a log versus

e.m.f. plot, where is the equivalents of current passed

for the production of the lower oxidation state. Initial

64

negative deviations from the linear portion of the plot were

assumed to be due to the presence of some of the reduced

species before its coulometric production began. The amount

of this initial "impurity" was calculated and corrections

were applied to the data.

The corrected values of the mn products are 2.0 + .1 for

cadmium, 1.0 + .1 for lead, and —3 for tin. The cadmium re­

sult concurs with previous studies which suggest that Cd^^ is

the reduced species of cadmium formed in various melts. How­

ever, mn = 1 for lead indicates that an ion of the type Pb+,

Pb2^, Pb]^, etc. is formed in Pb(AlCl4)2-NaAlCl4 (90 mole

percent)-AlCl3, while other workers have concluded that Pb^^

is formed in molten PbCl2. The mn product for tin suggests

one of the species Sn^, Sn^"*", Sn^*, etc., which is the first

determination of a specific reduced species for tin. Drift­

ing potentials prevented the designation of a value for zinc.

The study of nickel was not successful because NiCl2 is

insoluble in NaAlCl^.

Pb(AlCl^)2 and Sn(AlCl^)2 were synthesized. Procedures

were developed for preparing highly pure AICI3, NaAlCl^,

Pb(AlCl4)2 5 and Sn(AlCl4)2. Solubilities were determined for

lead and tin in their tetrachloroalumiriates and for cadmium,

65

lead, and tin in their tetrachloroaluminates (5 mole %) in

NaAlCl^ (90 mole %)-AlCl2 (5 mole %). The changes in these

solubilities caused by addition of NaAlCl^-AlClg to the pure

cadmium, lead, or tin tetrachloroaluminates are semi-

quantitatively consistent with the above formulations of

lower oxidation state species.

1

2

3

4

5

6

7

8

9

10

11

12

13

66

BIBLIOGRAPHY

Corbett, J. D., Burkhard, W. J., and Druding, L» F. J. Am. Chem. Soc., 76 (1961).

Corbett, J. D. and McMullan, R. K. J. Am. Chem. Soc., 77, 4271 (1955).

Corbett, J. D. and Hershaft, A. J. Am. Chem, Soc., 80, 1530 (1958).

Clark, R. J., Griswold, E., and Kleinberg, J. J. Am. Chem. Soc., 1532 (1958).

Hershaft, A. and Corbett, J. D, 551 (1962).

Hershaft, A. and Corbett, J. D. (1963).

Schumb, W. C. and Rogers, H. H. 5806 (1951).

Corbett, J. D. The solution of metals in their molten salts. In Sundheim, B. R., ed. Fused salts, pp. 341-407. McGraw-Hill, New York. (1964).

Corbett, J. D. Inorg. Chem., 1, 700 (1962).

Woodward, L. A., Garton, G., and Roberts, H. L. J. Chem. Soc., 3723 (1956).

Foster, L. M. , Russell, A. S., and Cochran, C. N. J. Am. Chem. Soc., 72., 2580 (1950).

Corbett, J. D. and Winbush, S. v. J. Am. Chem, Soc,, 77, 3964 (1955).

Palkin, A. P., Ostrikova, N. V., and Vigutova, T. N. Russ. J. Inorg. Chem., 8, 1344 (1963).

J. Chem. Phys., 36,

Inorg. Chem., 2, 979

J. Am. Chem. Soc,, 73,

67

14, Fedorov, P. I. and Fadeev, V. N. Zh, Neorgan. Khim., 9, 378 (1964). Original available but not translated; abstracted in Chem. Abstracts, 10194g (1964),

15, Fadeev, V. N. and Fedorov, P. I. Zh. Neorgan. Khim., 9, 381 (1964). Original available but not translated; abstracted in Chem. Abstracts, ll419d (1964),

16, Topol, L. E« J. Phys. Chem,, 67, 2222 (1963),

17, Gorodis'kii, 0. V. Dopovidi Akad. Nauk Ukr. R,S,R,, 995 (1959). Original not available; abstracted in Chem. Abstracts, 22102a (1960).

18, Gorodyskii, 0. V., Delimarskii, Yu. K,, and Grishchenko, V. F. Academy of Sciences of the USSR, Physical Chemistry Section, Proceedings, 150. 434 (1963),

19, Kolotii, A. A, Ukr. Khim, Zh,, 29, 939 (1963). Orig­inal not available; abstracted in Chem. Abstracts, 60, 6487b (1964).

20, Vitkin, A, I. and Plotnikova, T. P. Sb. Tr. Tsentr, Nauchn.-Issled Inst. Chernoi Met., No, £8, 146 (1962), Original not available; abstracted in Chem, Abstracts 58, 8644d (1963).

21, Karpachev, S., Stromberg, A., and Jordan, E, Zh, Fiz, Khim., 18, 43 (1944).

22, Egan, J. J. J. Phys, Chem,, 2222 (1961).

23, Corbett, J. D, and Albers, F. C. J, Am. Chem. Soc,, 533 (1960).

24, Bruner, B. L. and Corbett, J. D. J, Inorg. Nucl. Chem., 20, 62 (1961).

25, McMullan, R. K. and Corbett, J. D. J. Am. Chem. Soc., 4761 (1958).

26, Corbett, J. D. and McMullan, R. K. J, Am. Chem, Soc,, 78, 2906 (1956),

68

27. Topolj L. E., Yosim, S. J., and Osteryoung, R. A« J. Chem. Phys., 1511 (1961).

28. Flengas, S. N. and Ingraham, T. R. Can. J. Chem., 35. 1139 (1957).

29. Hill, D. L.5 Perano, J., and Osteryoung, R. A. J. Electrochem. Soc., 106. 698 (1960).

30. Senderoff, S. and Brenner, A. J. Electrochem. Soc., 101, 31 (1954).

31. Bockris, J. CM., Hills, G. J., Inman, D, , and Young, L. J. Sci. Instr., 438 (1956).

32. Laity, R. W. Electrodes in fused salt systems. In Ives, D. J. G. and Janz, G. J. Reference electrodes, pp. 524-606. Academic Press, New York (1961).

33. Kolotii, A. A. Ukr. Khim. Zh., 29, 1169 (1963). Original not available; abstracted in Chem. Abstracts, 60, 6487d (1964).

34. Fischer, V. W. and Simon, A. Z. anorg. allgem, Chem., 306. 1 (1960).

35. Naryskin, I. I. J, Phys. Chem. (U.S.S.R.), 13, 690 (1939). Original available but not translated; abstracted in Chem. Abstracts, M, 1529^ (1940).

36. Dewing, E. W. J. Am. Chem. Soc., 77., 2639 (1955).

37. Corbett, J. D., Winbush, S. v., and Albers, F. C. J. Am. Chem. Soc., 79, 3020 (1957).

38. Johnson, J. W., Cubicciotti, D., and Kelley, C. M. J. Phys. Chem., 1107 (1958).

39. Barber, M. , Linnett, J. W., and Taylor, N. H. J. Chem. Soc., 3323 (1961).

40. Hansen, M. and Anderko, K, Constitution of binary alloys. McGraw-Hill, New York (1958).

69

41. Delimarskii, lu» K. and Markov, B. F. Electrochemistry of fused salts. The Sigma Press, Washington, D.C. (1961).

42. Foster, L. M. J. Am. Chem. Soc., 7_2} 1902 (1950).

43. Morrey, J. R. Inorg. Chem., 2, 163 (1963).

44. Jander, V. G. and Swart, K. Z. anorg. allgem. Chem., 301. 54 (1959).

45c Karpachev, S. and Stromberg, A. Zh. Fiz, Khim. , _13, 397 (1939).

46. Ibers, J. A. Acta Cryst., _15, 967 (1962).

47. Delimarskii, lu. K., Skobets, E. M., and Berenblyum, L. S. Zh. Fiz. Khim., 22, 1108 (1948).

48. Vsrdisck, R. G. and Yntema, L. F. J. Phys, Chem,, 46, 344 (1942).

49. Gorenbein, E. Ya. and Kavet'skii, M. S. Russ. J. Phys. Chem., 37, 82 (1963).

50. Topol, L. E. and Osteryoung, R. A. J. Electrochem. Soc., 108, 573 (1961).

70

ACKNOWLEDGMENTS

The author sincerely appreciates the guidance and

stimulation provided by Dr. John D. Corbett, the friendship

and dialogue shared with Physical and Inorganic Chemistry

Group IX, especially Dick Barnes and Buddy Bruner, and the

encouragement of his wife, Betty.

71

APPENDIX

Table 5. X-ray powder diffraction data for Pb(AlCl^^2

o d(A) Relative intensity

o d(A) Relative intensity

7.02 100 3.13 40

6.32 10 3.06 10

6.15 40 2.97 30

5.03 5 2.69 20

4.82 20 2.26 10

4.55 70 2.24 10

3.62 20 1.69 5

3.28 20

72

Table 6« X-ray powder diffraction data for SnCl2

o o d(A) Relative intensity d(A) Relative intensity

4.57 20 2.20 20

3.95 30 2.17 10

3.54 100 2.11 50

2.95 10 1.92 5

2.92 5 1.77 5

2.78 30 1.70 5

2.52 40 1.58 5

2.29 5 1,42 5

2.25 10

73

Table 7, X-ray powder diffraction data for SnfAlCl^^g

o o d(A) Relative intensity d(A) Relative intensity

7.37 20 2.93 10

6.19 100 2.87 10

5.15 40 2.82 5

4.62 10 2.75 10

3.88 40 2.60 50

3.62 5 2.57 10

3.53 10 2.52 60

3.40 20 2.46 10

3.20 10 2.37 10

3.10 30 2.19 5

2.99 10 2.15 5